The all-new audience model empowers marketers with the ability to run targeted campaigns on non-ID inventory without compromising scale, quality or campaign results.
SAN FRANCISCO, Calif. (May 19, 2021) – Emodo, an Ericsson company focused on premium mobile advertising solutions, today announced the launch of Emodo Predictive Audiences, a new ML-powered targeting solution that enables advertisers to target audiences at scale – whether the users’ devices have IDs, or not. Predictive Audiences complement the company’s other targeting products to provide an accurate, comprehensive way forward as the ad industry moves away from identifiers and as consumer views on privacy shift.
Apple’s iOS 14.5 update officially launched on April 26th, severely limiting ad targeting across the platform’s mobile devices as the Identifier for Advertisers (IDFA) is now offered only on an opt-in basis at the user level. As a result, the industry has seen a reduction in audience for ID-based campaigns. This change has created a need for innovative, privacy-forward solutions for advertisers. As the industry continues a shift away from user-level targeting, machine learning models built on accurate truth sets are playing a significant role in targeting.
Emodo Predictive Audiences leverage segment-specific machine learning algorithms to classify bid requests into audience segments in real-time. The audience algorithms are trained by the industry’s most accurate, persistent truth sets – including fully opt-in mobility and usage data from mobile operators.
“One thing we learned from GDPR is that opt-in identity models, while effective, can lack sufficient reach for many brands,” remarked Emodo CEO, Alistair Goodman. “Machine learning models create privacy-friendly, intended audiences that are as efficient as traditional solutions and complement ID targeting without compromising scale, quality or campaign impact.”
When bids in the programmatic bid stream no longer have IDs associated with them, Emodo’s Predictive Audience algorithms identify the intended audience without IDs by using algorithms designed by evaluating hundreds of attributes in real time. Early uses of Emodo’s predictive audience segments have shown the ability to reach audience members at rates up to 10X more efficient than random targeting, with prediction accuracy rates that range from 70% to 90%.
For the last three years, Emodo has built and applied machine learning models to solve a variety of problems, from improving location accuracy to removing fraud/IVT. Enriching audience segmentation is a logical evolution which enables the company to protect consumer privacy while delivering the impact and scale that advertisers require.
With the launch of Predictive Audiences, Emodo Audience segments are now comprised of three primary targeting methods: Verified (ID-based), Predictive and Contextual targeting. The three methods complement each other and work in tandem to produce a wide range of addressable audience segments with significant scale. To ensure maximum compatibility and flexibility, Emodo is augmenting its proprietary offerings by implementing support for the industry’s most broadly adopted alternative ID structures.
Marketers interested in leveraging Emodo Audiences to improve ad